Calculating Receiver Sensitivity


According to the application note, APS017 section 5.3.1:
APS017_Max-Range-in-DW1000-Systems_v1.1.pdf (847.9 KB)

Receiver sensitivity is quoted in the DW1000 datasheet and depends on a number of parameters including the selected channel and the data rate. For longest range the lowest data rate (110 kbps) should be used. See [DW1000 Datasheet] and [DW1000 User Manual] for further information.

I can not find any more information in the datasheet or user manual to help with the calculation for receiver sensitivity. To my understanding, typical receiver noise floor is given by:
Noise floor = -174dBm/Hz + 10logBW + NF
Since BW = 500MHz and NF ~5dB, the noise floor should be about -82dBm/500MHz.

What other parameters should I consider to achieve a receiver sensitivity similar to the one described in the DW1000 datasheet (-106dBm/500MHz for 1% PER, 110 kbps, Preamble 2048)?

You can calculate the receive signal strength using the formulas in the user manual sections 4.7.1 and 4.7.2. That does not calculate the receiver sensitivity directly, but then you can attenuate your transmitter until you reach the edge of reception and see what the numbers are.

Note that the signal strength given by the formulas is not an exact fit, particularly at higher powers, see figure 22 in the user manual for the measured versus true curve.

In our testing, we have not found the 4.7.1 and 4.7.2 formulas to be a very good tool for sensitivity measurement. The readings seem not particularly stable and affected by environmental factors such as multipath. Instead, we have a transmitter with a computer controlled variable attenuator which we adjust up and down to find the relative sensitivity. That doesn’t report directly in dBm, but we can make very careful and repeatable relative measurements when we are testing possible antennas. Thus we can say antenna A is better than antenna B, or that board A is better than board B, but not give hard absolutes on the receiver dBm sensitivity.

Interference, both from external sources and from your own board itself (such as switching regulators, processor clocks, etc) can radically affect your reception sensitivity. Good power filtering and a shield on the DW1000 circuits helps, though these are not universally required depending on the design particulars.

You are not actually achieving 500 MHz of usable bandwidth, instead the processing gain of correlating the impulse to the incoming signal decreases the noise floor substantially since the noise is mostly uncorrelated to the impulse shape. This process trades channel bandwidth (now 6.81 Mbps at most) for a lower noise floor.

Once properly configured, the DW1000 should meet that datasheet spec though I can’t say we have verified that ourselves. Note that this figure is at the DW1000 pins, so how one measures this exactly is a bit tricky and is ultimately an academic exercise. The best approach would be to measure it at the antenna connector (antenna into power meter), and then adjust for expected path loss into DW1000.

Given that Decawave has already optimized the settings for best range, the most definitive way to get longer range is to add an LNA (low noise amplifier) to the circuit. This basically eliminates the effect of the DW1000 NF of ~5 dB. Our typical improvement is a little under double the range. Most of our custom UWB node designs this year have included an LNA for this reason.

Mike Ciholas, President, Ciholas, Inc
3700 Bell Road, Newburgh, IN 47630 USA
+1 812 962 9408